A New Region-Based Minimal Path Selection Algorithm for Crack Detection and Ground Truth Labeling Exploiting Gabor Filters

Round 1

Reviewer 1 Report

Some comments formulated during my review are presented below. These are as follows:

-I consider it is not a clear contribution to the aims of Remote Sensing journal. The topic involved in the paper is not sufficient for publication in this journal. In the work, much attention is paid to presenting the theory, based on other works, and to a much lesser extent to the issues of remote sensing image segmentation. Nevertheless, the Authors made no effort whatsoever to make the manuscript more relevant to readers of Remote Sensing. For an overview of the Aims & Scope, please have a look at the journals’ homepage. 

-The topic of crack detection is very popular and there are a lot of papers on this topic. Unfortunately, in the description of the experimental part there is no comparison to the recent state-of-the-art methods. And this is a very serious limitation of this work. The research results reported are too premature for publication. More work is needed to substantiate the conclusions in your manuscript.

 

Some comments formulated during my review are presented below. These are as follows:

-I consider it is not a clear contribution to the aims of Remote Sensing journal. The topic involved in the paper is not sufficient for publication in this journal. In the work, much attention is paid to presenting the theory, based on other works, and to a much lesser extent to the issues of remote sensing image segmentation. Nevertheless, the Authors made no effort whatsoever to make the manuscript more relevant to readers of Remote Sensing. For an overview of the Aims & Scope, please have a look at the journals’ homepage. 

-The topic of crack detection is very popular and there are a lot of papers on this topic. Unfortunately, in the description of the experimental part there is no comparison to the recent state-of-the-art methods. And this is a very serious limitation of this work. The research results reported are too premature for publication. More work is needed to substantiate the conclusions in your manuscript.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper proposes a region-based minimum path selection algorithm that utilizes Gabor filters for crack detection and conducts extensive research on the minimum.

The method is innovative and the workload is sufficient. But in my opinion, there are still the following parts of the paper that need to be corrected:

 

1. Lines 62-79 are poorly written and illogical. It is recommended to modify the content.

2. The contribution part is messy and simple, which should be rewritten.

3. The descriptions of the letters in many formulas are insufficient and need to be supplemented.

4. The descriptions of Figure 5, Figure 7, Figure 32, Figure 33, and Figure 34 are insufficient. The definition of all parameters on the abscissa and the description of this group of figures should be supplemented.

5. Line 318, not the picture 'on the left' but the picture in the center.

6. For the evaluation indicators TP, TN, FP, and FN, calculation formulas need to be supplemented.

7. The content of lines 383-385 is redundant.

8. The content of the conclusion needs to be revised, which is not concise, and most of the background-related introduction should be deleted. In this part, The contribution of the algorithm is the one that needs the most attention.

9. Lines 193-199, the description of the experimental data should be a separate section. It is recommended to write some cohesive statements about the proposed algorithm in this section.

Part of the English expression of the paper is not fluent and unnatural. The paper is suggested to be revised by native speakers to improve the quality of the expressions.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The present study details the methodology for a new algorithm for crack segmentation based on the theory of minimal path selection combined with a region-based approach obtained through the segmentation of texture features extracted using Gabor filters. A pre-processing step is described enabling to equalization of brightness and shadows which results in better detection of local minima. The paper is well written and I would like to give several minor suggestions.

(1) At the thematic level, the proposal provides a very interesting vision, as the "black-box" system of deep learning does make it difficult for researchers to explain the underlying principles based on the results. Therefore, traditional image processing that can track intermediate processes is used for crack detection. However, the effect of traditional image processing is better than that obtained by current deep learning, which is an important limitation concerning the aspirations of the proposal. These limitations should be assumed with more rigor and realism in the development of the argumentation of the manuscript.

(2) The authors may add more state-of-art application articles for the integrity of the manuscript.

(3) For visual measurement applications, please refer to Novel visual crack width measurement based on backbone double-scale features for improved detection automation;. automatic classification of asphalt pavement cracks using a novel integrated generative adversarial networks and improved VGG model.

(4) Try to place the relevant pictures or tables mentioned in the paper on adjacent pages to make it easier for readers to read. For example, Figure 7 was mentioned on page 7, but Figure 7 is on page 11.

(5) There is a mark in Figure 34, D, and it can be seen from C that the mark has been identified as a spiral crack. I suggest searching for a crack image in the database or in reality to redo the detection results to improve the feasibility of your data.

(6) I recommend comparing DC with RB-MPS. This part should provide more figure comparison experiments rather than just a table of results for comparison. Also, you should provide more evidence to prove why the F1 score of DC is higher than RB-MPS, but you choose RB-MPS. The conclusion that RB-MPS is better than DC in this situation is not convincing enough.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

1. For question 2, the intended meaning of 'The contribution' is in reference to the contribution of the work and method, rather than the authors. In your article, it appears as 'The main improvements in the new algorithm'.

2. The conclusion could benefit from a more concise presentation. While a brief introduction to the topic is necessary, it may be advantageous to reduce the length of the background description. Even when reading the paper, readers will likely be more interested in the proposed algorithm. To this end, I suggest consolidating the first and second paragraphs, and shortening the length accordingly.

I think the quality of English is generally good. However, I did notice some areas where the language could be improved to sound more natural and authentic. Specifically, there were instances of phrasing that did not quite reflect how native speakers would typically express the same ideas.

Author Response

Please see the attachment.

Author Response File: Author Response.docx